Optimization of Microenvironments Inducing Differentiation of Tonsil-Derived Mesenchymal Stem Cells into Endothelial Cell-Like Cells

Abstract

Background:

Stem cell engineering is appealing consideration for regenerating damaged endothelial cells (ECs) because stem cells can differentiate into EC-like cells. In this study, we demonstrate that tonsil-derived mesenchymal stem cells (TMSCs) can differentiate into EC-like cells under optimal physiochemical microenvironments.

Methods:

TMSCs were preconditioned with Dulbecco’s Modified Eagle Medium (DMEM) or EC growth medium (EGM) for 4 days and then replating them on Matrigel to observe the formation of a capillary-like network under light microscope. Microarray, quantitative real time polymerase chain reaction, Western blotting and immunofluorescence analyses were used to evaluate the expression of gene and protein of EC-related markers.

Results:

Preconditioning TMSCs in EGM for 4 days and then replating them on Matrigel induced the formation of a capillary-like network in 3 h, but TMSCs preconditioned with DMEM did not form such a network. Genome analyses confirmed that EGM preconditioning significantly affected the expression of genes related to angiogenesis, blood vessel morphogenesis and development, and vascular development. Western blot analyses revealed that EGM preconditioning with gelatin coating induced the expression of endothelial nitric oxide synthase (eNOS), a mature EC-specific marker, as well as phosphorylated Akt at serine 473, a signaling molecule related to eNOS activation. Gelatin-coating during EGM preconditioning further enhanced the stability of the capillary-like network, and also resulted in the network more closely resembled to those observed in human umbilical vein endothelial cells.

Conclusion:

This study suggests that under specific conditions, i.e., EGM preconditioning with gelatin coating for 4 days followed by Matrigel, TMSCs could be a source of generating endothelial cells for treating vascular dysfunction.

Introduction

Cardiovascular diseases (CVDs) are the most common cause of death worldwide. The World Health Organization estimates that approximately 27% of all global deaths were caused by ischemic heart disease and stroke in 2016, designating them as ‘the world’s biggest killers’ for the past 15 years [1]. Endothelial dysfunction, pathological conditions with impaired vasorelaxation, and decreased nitric oxide (NO) production in blood vessels are well-defined predictors of CVDs such as atherosclerosis [2]. Therapies or medications that reduce hypertension/blood pressure have generally been used as treatments for endothelial dysfunction [3]. Recent clinical studies have looked at ways to upregulate endothelial NO synthase (eNOS) activity to increase the bioavailability of NO [4].

Regenerative tissue engineering of blood vessels is an attractive therapeutic approach because it provides alternative endothelial cell (EC)-like cells that could potentially restore or improve the function of damaged ECs [5]. Current methods for in vitro pre-vascularization using ECs and endothelial progenitor cells (EPCs) have been suggested, but the lack of expansion capacity and vascularization are yet to be overcome [6]. Sethe et al. (2006) and Son (2017) reviewed that MSCs unlike embryonic stem cells have limited proliferation capacity due to in vitro aging, leading to cellular senescence and losing differentiation potentials [7, 8]. TMSCs have relatively higher proliferation rate compared to MSCs derived from bone marrow and fat, and their differentiation potentials remain to be stable till passage 15 [9]. Some researchers have used cytokine treatments such as vascular endothelial growth factor (VEGF) to differentiate mesenchymal stem cells (MSCs) into ECs, but they have had limited success. Furthermore, MSCs were also reported to have the potential to differentiate into cells that express early markers of vasculogenesis [10–12]. Nonetheless, most available MSCs face issues such as limited cell availability, proliferation capacity, and an invasive protocol for obtaining target cells including ECs.

For the past several years, we have been using tonsil-derived MSCs (TMSCs) for various stem cell research applications. TMSCs can easily be obtained from the discarded tonsil tissue of children undergoing tonsillectomy, and they are readily available due to their high proliferation rate as well as low immunogenicity and tumorigenicity [9, 13, 14]. Furthermore, they have great differentiation potential into various cell types, including osteoblasts, adipocytes, myocytes, and parathyroid hormone- and insulin-releasing cells [9, 15–17], which has made them an appealing MSC source for regenerative tissue engineering.

We previously reported that TMSCs produced CCN1 [18], which is known to promote the differentiation of EPCs and reendothelialization [19]. Therefore, in this study, we further investigated the potential efficiency of differentiating TMSCs into EC-like cells. Previously, certain factors, such as culture medium compositions/supplements and the physical properties of culturing conditioning [20], were shown to induce efficiency in EC differentiation. Among them, gelatin is often used to improve the attachment and growth of ECs [21, 22], and it has generally been used for culturing specific type of ECs such as human umbilical vein ECs (HUVECs) [23, 24]. Our study demonstrates that TMSCs under appropriate physiochemical microenvironments, such as preconditioning TMSCs in endothelial growth medium (EGM) on gelatin coating, can differentiate into EC-like cells, which could potentially be used for regenerative engineering in blood vessel research.

Materials and methods

Materials

DNase I and gelatin (G9391) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Collagenase type I, EGM, fetal bovine serum (FBS), newborn calf serum, low serum growth supplements (LSGS), antibiotic–antimycotic solution, trypsin, ethylenediaminetetraacetic acid (EDTA) solution, oligo (dT)₁₅, M-MLV reverse transcriptase, and SuperSignal™ West Femto Chemiluminescent Substrate were purchased from Invitrogen (Thermo, Waltham, MA, USA). Ficoll-Paque and enhanced chemiluminescence (ECL) reagents kit were purchased from GE Healthcare (Piscataway, NJ, USA). High-glucose (4500 mg/l) Dulbecco’s Modified Eagle Medium (DMEM-HG) and Dulbecco’s phosphate buffered saline (PBS) were purchased from Welgene Inc. (Gyeongsan, Korea). An EGM Bullet Kit™ was purchased from Lonza Bio (Portsmouth, NH, USA). Recombinant human cellular communication network factor 1 (rhCCN1) and VEGF were obtained from Cell Sciences (Canton, MA, USA) and R&D Systems (Minneapolis, MN, USA), respectively. Protease inhibitor mixture was purchased from Roche Applied Science (Penzberg, Germany). QIAzol lysis reagent was purchased from Qiagen (Hilden, Germany). An Illumina TruSeq RNA library kit was purchased from Illumina Inc. (San Diego, CA, USA). Antibodies for protein kinase B (Akt), p-Akt-Ser⁴⁷³, and CCN1 were purchased from Cell Signaling Technology (Beverly, MA, USA). Antibody for eNOS was obtained from Transduction Laboratories (Lexington, KY, USA). Antibody for von Willebrand factor (vWF) was purchased from Abcam (Cambridge, MA, USA). 4′,6-Diamidine-2′-phenylindole dihydrochloride (DAPI) was purchased from Calbiochem (San Diego, CA, USA). Matrigel was obtained from BD Biosciences (Bedford, MA, USA). Sterile nylon plastic cell strainers and all cell culture plates/flasks used in this study were obtained from Corning Costar (Cambridge, MA, USA). All other chemicals were of the purest analytical grade.

TMSC isolation, culture, and treatments

Tonsillar tissues were obtained from tonsillectomies performed at the Department of Otorhinolaryngology–Head and Neck Surgery, Ewha Womans University Medical Center (Seoul, Korea), with informed written consent. This experimental protocol was approved by the Institutional Review Board of Ewha Womans University Medical Center (IRB No. ECT-11-53-02).

TMSCs were isolated from the tonsillar tissues as described in our previous studies [9, 25]. In brief, tonsillar tissues were collected from 3 donors who received tonsillectomies (2 boys and 1 girl, < 10 years of age). The isolated tissues were mechanically digested by cutting and mincing and then enzymatically digested with collagenase type I and DNase I at 37 °C for 30 min. The digested tissues were sequentially filtered through wire mesh, 100 and 70 μm of nylon plastic strainers. The mononuclear cells were isolated using Ficoll-Paque density gradient centrifugation, creating white buffy coat within the gradient. The buffy coat was collected, and the cells were washed with PBS. The isolated cells were then cultured in DMEM-HG supplemented with 10% FBS and antibiotic–antimycotic, and allowed to adhere to the culture plates. The attached mononuclear cells were taken as TMSCs, and they were expanded and stored in liquid nitrogen until use. The surface markers of TMSCs and their mesodermal differentiation potentials have already been characterized and provided in the previous paper [26]. TMSCs of passages 5–7 were used for the experiments. All experiments were conducted in accordance with the approved guidelines and regulations.

For EC differentiation, 2 × 10⁵ TMSCs were preconditioned with DMEM supplemented with 10 or 50 ng/ml of VEGF. After 7 days, the cells were detached and recultured on Matrigel in DMEM for 10 days to induce the formation of a capillary-like network, which is one of phenotypical characteristics of EC-like cells. However, this did not induce capillary-like network formation in the TMSCs, so we modified the protocol as described below.

TMSCs were preconditioned in DMEM or EGM supplemented with 10% FBS for 4 or 7 days without incorporating VEGF followed by reculturing them on Matrigel with their respective media for up to 10 days. The protocol with 4 days of preconditioning successfully produced a capillary-like network, as shown in Fig. 1. Therefore, we further adjusted the conditions to enhance its integrity. After 4 days of preconditioning in EGM, the TMSCs were recultured on the Matrigel-coated surface in EGM supplemented with either 10 or 50 ng/ml of VEGF, 0.1 µg/ml of rhCCN1 or conditional medium from HUVECs. In separate experiments, the TMSCs were preconditioned on 0.2% gelatin-coated surface in DMEM or EGM for 4 days, followed by reculturing on Matrigel in the same respective medium with or without the addition of rhCCN1. The gelatin-coated plates were prepared by adding 10 µl of 0.2% liquefied gelatin solution (sterilized by autoclaving at 121 °C, 15 psi for 30 min.) followed by drying for 2 h at 37 °C. The excessive liquid was removed by suction prior to use. For comparison, HUVECs were used as a positive control.

Fig. 1.

Light-microscope view (×100) of morphological changes during endothelial differentiation of TMSCs that were preconditioned in DMEM or EGM for 7 days (−7 d) followed by replating on Matrigel (2 × 10⁵ cells/500 μl Matrigel) with the corresponding media. The pictures were taken at 1, 4, 7, and 10 days, and a cotton thread-like structure was observed

Isolation of HUVECs and preparation of conditioned medium

HUVECs were isolated with the written informed consent of each woman who donated an umbilical cord. The Institutional Review Board of the Ewha Womans University Mokdong Hospital approved this study (IRB No. ECT-11-53-02).

In brief, HUVECs were isolated from fresh newborn umbilical cord veins by collagenase digestion using the standard protocol, as previously described [27]. Isolated cells were cultured on plates coated with 0.2% gelatin in M200 supplemented with LSGS, 10% FBS, and antibiotics at 37 °C under 5% CO₂. HUVECs between passages 5 and 7 were used for all experiments. Conditioned medium collected from HUVECs (HUVEC-CM) was acquired from the culture medium, as reported previously [18]. Briefly, cultured medium for HUVECs (80% confluency) was collected and concentrated by freeze-drying, resuspended with PBS, and stored at − 80 °C until use. The concentration of CM was determined using a BCA protein quantitation assay (Sigma-Aldrich, St. Louis, MO, USA), and final concentrations of 0.05 and 0.1 μg/ml (previously determined from an MTT assay, data not shown) were used for trial.

Capillary-like structure formation assay

EC-like cell differentiation was assessed using microscopy by observing the formation of a capillary-like network on Matrigel in a 24-well plate, and those observations were processed using the IMT(VT)-size image analysis program (iMTechnology, Suwon, Korea).

In this study, HUVECs were used as a positive control to evaluate the EC-like cell differentiation potential of TMSCs. In vitro tube formation was carried out as previously described [28]. Briefly, cultured HUVECs were seeded on a 24-well Matrigel-coated culture plate, and capillary-like network formation was observed up to 20 h under a phase-contrast microscope.

Sample preparation for microarray and quantitative real time polymerase chain reaction (qRT-PCR)

The differential effects of EGM and DMEM preconditioning on the EC-like cell differentiation of TMSCs were investigated by performing a microarray followed by qRT-PCR. The TMSCs were preconditioned in either DMEM or EGM for 4 days. RNA was extracted from the TMSCs using QIAzol lysis reagent and column purified with an RNeasy mini kit (Qiagen, Hilden, Germany). The purified RNA was treated with DNase I to remove genomic DNA. The RNA concentration and integrity of each sample were measured using an Agilent 2100 Bioanalyzer (Agilent Technology, Santa Clara, CA, USA). Those RNA samples with an RNA integrity number ≥ 8 were used for the microarray analyses and qRT-PCR.

Sequence annotation and identification of differentially expressed genes (DEGs)

For the microarray analyses, cDNA libraries was constructed using the GeneChip WT (Whole Transcript) Amplification kit as described by the manufacturer. The sense cDNA was then fragmented and biotin-labeled with TdT (terminal deoxynucleotidyl transferase) using the GeneChip WT Terminal labeling kit. Approximately 5.5 μg of labeled DNA target was hybridized to the Affymetrix GeneChip Array at 45 °C for 16 h. Hybridized arrays were washed and stained on a GeneChip Fluidics Station 450 and scanned on a GCS3000 Scanner (Affymetrix, Santa Clara, CA, USA). Array data export processing and analysis was performed using Affymetrix^® GeneChip Command Console^® Software (AGCC). For the DEG analysis, the log2 values (FPKM + 1) were calculated and then normalized by quantile. Transcripts with fold-change values larger than 2 with a p value ≤ 0.05 were included in the analysis as DEGs. A hierarchical clustering analysis was performed using complete linkage and Euclidean distance as the measure of similarity to display the DEG patterns. All DEG data analysis was conducted using Affymetrix^® Expression Console™ Software R 3.1.2 (www.r-project.org).

Gene ontology (GO) and enrichment analysis

Functional groups and pathways encompassing the DEGs were identified using GO and a Kyoto Encyclopedia of Genes and Genomes pathway analysis in the Database for Annotation, Visualization, and Integrated Discovery software (DAVID v.6.8, Frederick, MD, USA). The threshold was set as a modified Fisher exact p value (EASE score) ≤ 0.05.

Quantitative real time polymerase chain reaction

For the qRT-PCR analyses, first-strand cDNA was synthesized with 1 µg of RNA, oligo (dT)₁₅ (Thermo Fisher Scientific, Waltham, MA, USA), and M-MLV reverse transcriptase (Thermo Fisher Scientific, Waltham, MA, USA).

Quantitative PCR amplification of the cDNA encoding for hypoxia inducible factor 1α (HIF1α), interleukin-8 (IL8), angiopoietin-1 (ANGPT1), epiregulin (EREG), and ectonucleotide pyrophosphatase/phosphodiesterase 2 (ENPP2) was carried out using rTaq DNA polymerase (Thermo Fisher Scientific, Waltham, MA, USA) and QuantStudio 3 (Applied Biosystems, CA, USA). Primer sequences and product sizes are provided in Table 1.

Table 1.

Summary of the designed oligonucleotide primer sequences (F: forward primer, R: reverse primer)

Gene	Primer sequence (5′–3′)	PCR product size (bp)
HIF1α*	F: GAA CGT CGA AAA GAA AAG TCT CG	124
	R: CCT TAT CAA GAT CGC AAC TCA CA
IL8	F: ACT GAG AGT GAT TGA GAG TGG AC	112
	R: AAC CCT CTG CAC CCA GTT TTC
ANGPT1	F: TCG TGA GAG TAC GAC AGA CCA	173
	R: TCT CCG ACT TCA TGT TTT CCA C
EREG	F: GTG ATT CCA TCA TGT ATC CCA GG	120
	R: GCC ATT CAT GTC AGA GCT ACA CT
ENPP2	F: ACT TTT GCC GTT GGA GTC AAT	102
	R: GGA GTC TGA TAG CAC TGT AGG A
IGFBP5	F: ACC TGA GAT GAG ACA GGA GTC	136
	R: GTA GAA TCC TTT GCG GTC ACA A
GAPDH	F: TAA CTT CTG TGC TGT GCC AGC C	103
	R: TTA AAA GCA GCC CTG GTG ACC

*HIF1α (hypoxia inducible factor 1α), IL8 (interleukin 8), ANGPT1 (angiopoietin 1), EREG (epiregulin), NR4A3 (nuclear receptor subfamily 4 group A member 3), ENPP2 (ectonucleotide pyrophosphatase/phosphodiesterase 2), IGFBP5 (insulin-like growth factor binding protein 5), GAPDH (glyceraldehyde-3-phosphate dehydrogenase)

Western blot analysis

Total protein was extracted from TMSCs using cell lysis buffer (20 mM Tris–HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 mM Ethylene Glycol Tetraacetic Acid 1 mM phenylmethylsulfonyl fluoride, 10 mM β-glycerophosphate, 1 mM NaF, and 1 mM Na₃VO₄) containing protease inhibitor mixture. Equal amounts of protein (30 µg) were separated by 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred onto nitrocellulose membranes. Blots were blocked with 5% skim milk for 1 h and then incubated with appropriate antibodies, followed by corresponding secondary antibodies and development using ECL reagents. Lysates from HUVECs (5 µg) were used as a reference, and β-Actin was used as the loading control.

Immunofluorescence assay (IFA)

At 10 h after reculturing the preconditioned TMSCs on Matrigel, the differentiated TMSCs were washed three times with PBS, and fixed using 4% paraformaldehyde in PBS for 10 min at room temperature. The cells were then permeabilized using 0.1% Triton X-100 in PBS for 10 min and washed with PBS prior to stain with vWF antibody (red). DAPI was used to stain nucleus of TMSCs (blue).

Statistical analysis

All data are presented as the mean ± standard deviation (SD). Statistical analyses were performed using Statistical Package for Social Sciences version 21 (International Business Machines Corporation, Armonk, NY, USA). Statistical significance between the two experimental groups, DMEM and EGM, was analyzed using the Student’s t-test, and (*) and (**) denote significant differences at p < 0.05 and 0.01, respectively.

Results

Formation of the capillary-like structure by TMSCs in the EC growth medium

The supplementation of VEGF during preconditioning was previously shown to induce EC-like cell differentiation in human MSCs, promoting the formation of a capillary-like network upon culturing on Matrigel [29, 30]. In the present study, however, we could not find any signs of capillary-like network formation in TMSCs under the conditions used by other authors, preconditioning TMSCs in DMEM containing 10 or 50 ng/ml VEGF. When we modified the protocol, eliminating the VEGF during preconditioning, the TMSCs aggregated as a cotton thread-like structure 10 days after reculturing on Matrigel, but they had no apparent tubules or capillaries (Fig. 1, upper panel). When EGM was used in replacement of DMEM for preconditioning TMSCs for 7 days, the time required to form the cotton thread-like structure was shortened by 3 days; they started to appear 7 days after being replated on Matrigel (Fig. 1, lower panel).

When the preconditioning period was reduced to 4 days, the TMSCs preconditioned in DMEM aggregated much more quickly than those preconditioned in DMEM for 7 days, forming the aggregation of cotton thread-like structure within 4 h after replating on Matrigel (Fig. 2A, upper panel). The structure was maintained for 3 h, but it had completely dissipated 7 h after reculturing on the Matrigel. On the contrary, we observed the formation of a clear capillary-like network in the TMSCs preconditioned in EGM for 4 days, and it lasted longer than 3 h (Fig. 2A, lower panel). The capillary-like structure formed by the TMSCs preconditioned in EGM for 4 days was morphologically similar to those observed in HUVECs (Fig. 2B), but the HUVECs maintained the capillary-like network for much longer period of time, lasting ~ 20 h after replating on Matrigel.

Fig. 2.

Light-microscope view (×100) of morphological changes during endothelial differentiation of TMSCs that were preconditioned in DMEM or EGM for 4 days (− 4 d) followed by replating on Matrigel (2 × 10⁵ cells/500 μl Matrigel) with the corresponding media. A The pictures were taken at 4, 7, and 10 h to observe the formation of a capillary-like structure. B HUVECs were cultured on Matrigel with EGM and observed at 4, 7, 10, and 20 h for the formation of a capillary-like structure and used as a control. Arrowheads indicate capillary-like structure

DEG profiles between the TMSCs preconditioned in EGM and DMEM

Given that preconditioning TMSCs in EGM for 4 days successfully promoted the formation of a capillary-like structure similar to those in HUVECs, we investigated how EGM affected EC-like cell differentiation in TMSCs. To this end, we created a heat map by filtering DEG probes of TMSCs from the two different preconditioning culture media, EGM and DMEM (Fig. 3). In the GO clustering analysis, highly differentially regulated genes between TMSCs preconditioned with EGM and DMEM that were shared by all three donors were in the following categories: angiogenesis, blood vessel morphogenesis and development, and vascular development. The significant DEGs, genes that were upregulated or downregulated by more than 2-fold on average, are listed in Tables 2 and 3, respectively.

Fig. 3.

Heat map from microarray analyses showing differentially expressed genes (DEGs) between TMSCs with 4 days of preconditioning in two different media: EGM and DMEM

Table 2.

List of genes that were consistently upregulated in TMSCs preconditioned in EGM compared with those in DMEM. The genes upregulated by an average greater than twofold are listed here

Gene	Donor #1	Donor #2	Donor #3	Mean	SD
HIF1αa*	3.9431	9.3335	4.5059	5.9275	2.9631
EREGa	3.3772	9.1819	2.1822	4.9138	3.7443
IL8a	3.4290	7.3974	3.5759	4.8008	2.2499
NR4A3	3.6969	4.2275	3.0273	3.6506	0.6015
ENPP2a	4.5543	2.1825	2.3672	3.0346	1.3193
ANGPT1a	2.0429	4.5971	1.4750	2.7050	1.6631
EPGN	2.3757	3.5605	1.2090	2.3817	1.1758
GLMN	1.8817	3.2404	1.8379	2.3200	0.7974
EDNRA	1.5886	2.8502	2.0864	2.1751	0.6355
FOXF1	2.8341	2.3039	1.2584	2.1321	0.8018
MMP19	2.1620	2.1832	1.8917	2.0790	0.1625

*HIF1α (hypoxia inducible factor 1α), EREG (epiregulin), IL8 (interleukin 8), NR4A3 (nuclear receptor subfamily 4 group A member 3), ENPP2 (ectonucleotide pyrophosphatase/phosphodiesterase 2), ANGPT1 (angiopoietin 1), EPGN (epithelial mitogen), GLMN (glomulin), EDNRA (endothelin receptor type A), FOXF1 (forkhead box F1), MMP19 (matrix metallopeptidase 19)

^aGenes selected for PCR validation of the genome analysis

Table 3.

List of genes that were consistently downregulated in TMSCs preconditioned in EGM compared with those in DMEM. The genes downregulated by an average greater than twofold are listed here

Gene	Donor #1	Donor #2	Donor #3	Mean	SD
IGFBP5a*	− 5.8507	− 3.8086	− 2.7009	− 4.1201	1.5978
CXCL12	− 3.4099	− 1.6558	− 2.4822	− 2.5159	0.8775
BMP4	− 2.5162	− 1.5441	− 2.2276	− 2.0960	0.4992
EDN1	− 2.4616	− 1.4403	− 2.6110	− 2.1710	0.6372

*CXCL21 (C-X-C motif chemokine ligand 12), IGFBP5 (insulin-like growth factor binding protein 5), BMP4 (bone morphogenetic protein 4), EDN1 (endothelin 1)

^aGenes selected for PCR validation of the genome analyses

Among the genes listed in Tables 2 and 3, we selected the genes with the greatest fold difference or the genes that are closely associated with vasculature development and angiogenesis during blood vessel formation and validated their expression patterns using qRT-PCR (Fig. 4). Similar to the results from the microarray analyses, the PCR analyses confirmed that the TMSCs preconditioned with EGM had significantly higher gene expression of HIF1α (p < 0.01), IL8 (p < 0.05), ANGPT1 (p < 0.05), and ENPP2 (p < 0.05). The expression of EREG was also higher in the TMSCs preconditioned with EGM, but the difference was not significant because of high variability in the qRT-PCR data.

Fig. 4.

Quantitative RT-PCR analyses of TMSCs with 4 days of preconditioning in EGM or DMEM to validate the genome analyses from the microarray (hypoxia inducible factor 1α, HIF1α; interleukin 8, IL8; angiopoietin 1, ANGPT1; epiregulin, EREG; nuclear receptor subfamily 4 group A member 3, NR4A3; ectonucleotide pyrophosphatase/phosphodiesterase 2, ENPP2; insulin-like growth factor binding protein 5, IGFBP5). Statistical significance between the two experimental groups, DMEM and EGM, was analyzed using the Student’s t-test, and (*) and (**) denote significant differences at p < 0.05 and 0.01, respectively

As observed in the microarray analyses, the gene expression of insulin-like growth factor binding protein 5 (IGFBP5; p < 0.01) was significantly downregulated in TMSCs preconditioned with EGM compared to those with DMEM.

Effect of adding angiogenic cytokines and HUVEC-CM to the Matrigel on the formation of a capillary-like network by TMSCs

Our initial study revealed that preconditioning TMSCs with VEGF did not promote the formation of a capillary-like network. Given that VEGF is known to induce EC differentiation in bone marrow-derived MSCs (BM-MSCs) [30] and adipose-derived MSCs [31], we expected that VEGF exposure after the preconditioning period could promote the formation of a capillary-like network in TMSCs. To test this hypothesis, VEGF were supplemented to the EGM while reculturing preconditioned TMSCs on Matrigel to see whether VEGF improved the formation of a capillary-like network. However, VEGF supplementation (10 and 50 ng/µl) did not improve the network formation by TMSCs compared with EGM alone (Fig. 5A and B).

Fig. 5.

Light-microscope view (×100) of morphological changes during endothelial differentiation of TMSCs preconditioned in DMEM or EGM for 4 days followed by replating on Matrigel using the corresponding media with various conditional factors (2 × 10⁵ cells/500 μl Matrigel). A TMSCs without any supplementation. B TMSCs supplemented with 10 or 50 ng/µl of VEGF. C TMSCs supplemented with 0.1 µg/ml of rhCCN1. D TMSCs supplemented with 0.05 or 0.1 μg/ml of HUVEC-CM. E HUVECs with EGM and Matrigel. Arrowheads indicate capillary-like structure

In our next trial, we tested other known angiogenic cytokines/factors: rhCCN1 and HUVEC-CM. It was previously reported that an autologous source of CCN1 had angiogenic potential in TMSCs [18], and HUVEC-CM was previously shown to enhance the capillary-like network [32]. Nevertheless, we did not detect any synergistic angiogenic effect from rhCCN1 (Fig. 5C), and HUVEC-CM destabilized capillary-like network formation by TMSCs (Fig. 5D).

Gelatin coating in the preconditioning step improves the stability of the capillary-like network formed by the TMSCs

We found that angiogenic cytokines/factors played no significant role in improving the formation of a capillary-like network. Given that reculturing with Matrigel is essential for TMSCs to construct capillary-like networks, we wondered whether an appropriate physical microenvironment such as providing surface coating could improve capillary-like network formation by TMSCs. To this point, we found that the conditions used for HUVECs produced the most stable capillary-like network in this study (Figs. 2B and 5E); in this study, TMSCs were preconditioned on 0.2% gelatin-coated culture plates, a general condition used for culturing HUVECs [33].

The network established by the TMSCs preconditioned with gelatin were more stable than those without the gelatin coating (Fig. 2A), and the network structure was maintained for a longer period, lasting up to 10 h, compared with no gelatin incorporation (Fig. 6). As expected, no distinct capillary-like structure was observed in TMSCs preconditioned with DMEM irrespective of the gelatin coating, indicating that preconditioning with EGM is mandatory for TMSCs to establish a capillary-like structure. Supplementation with rhCCN1 in EGM or DMEM with the gelatin coating did not appear to significantly affect the stability or the time required for TMSCs to form capillary-like structures (Fig. 6).

Fig. 6.

Light-microscope view (×100) of morphological changes during endothelial differentiation of TMSCs preconditioned in DMEM or EGM with gelatin for 4 days followed by replating on Matrigel with or without rhCCN1 (0.1 µg/ml) in the corresponding media. The pictures were taken at 4, 7, and 10 h to observe the formation of capillary-like structures. Arrowheads indicate capillary-like structure

EGM preconditioning of TMSCs induces the expression of eNOS and other EC-related markers

Western blotting analyses revealed that weak but clear expression of EC-related marker, eNOS, occurred only in the TMSCs preconditioned in EGM followed by reculturing on Matrigel for 10 h (Fig. 7A). Virtually no eNOS expression was found in the TMSCs preconditioned in DMEM (Fig. 7A). IFA also confirmed that the TMSCs preconditioned in EGM with gelatin followed by reculturing on Matrigel for 10 h had much higher expression of vWF compared with those with DMEM (Fig. 7B), validating our results from the Western blotting analyses.

Fig. 7.

A Western blotting analyses of eNOS, p-Akt-Ser⁴⁷³, and total Akt from the protein lysates of TMSCs preconditioned in DMEM or EGM without gelatin for 4 days followed by replating on Matrigel (C), TMSCs preconditioned with gelatin coating followed by replating on Matrigel (G), and TMSCs preconditioned with gelatin followed by replating on Matrigel with rhCCN1 (G-rhCCN1). The TMSC lysates were collected at 10 h after the reculturing on Matrigel. Lysates from HUVECs (5 µg) were used as a reference, and β-Actin was used as the loading control. B Immunofluorescence assay (IFA) was performed to stain von Willebrand factor (vWF) from the TMSCs preconditioned either in DMEM or EGM with gelatin for 4 days followed by replating on Matrigel for 10 h. Blue, DAPI; Red, vWF

We also found higher expression of phosphorylated Akt at serine 473 (p-Akt-Ser⁴⁷³) an eNOS activity-related signaling factor, without a change in the expression of total Akt in the TMSCs preconditioned in EGM compared with those in DMEM. No p-Akt-Ser⁴⁷³ expression was observed in the basal HUVECs.

In contrast to the enhanced stability of the capillary-like network induced by gelatin (Fig. 6), neither gelatin alone nor gelatin with rhCCN1 significantly affected the expression of eNOS or p-Akt-Ser⁴⁷³ in TMSCs preconditioned in DMEM or EGM (Fig. 7A).

Discussion

Blood vessel regeneration using stem cells is a potential treatment that could mitigate endothelial dysfunction, thereby reducing the risk of CVDs. Oswald et al. (2004) were the first group to discover the potential of CD105⁺ CD73⁺ BM-MSCs to differentiate into EC-like cells in the presence of VEGF [30]. Subsequently, others have reported that MSCs derived from fat [29, 34], perinatal umbilical cord [29, 35], and placental chorionic villi [29] have a capacity to differentiate into EC-like cells, establishing a capillary-like network when placed on Matrigel.

In this study, we investigated the EC-like cell differentiation potential of TMSCs in different EC culture conditions, such as using EGM or DMEM and providing a gelatin surface coating during preconditioning. The TMSCs preconditioned in EGM established a capillary-like network and produced increased expression of EC-related markers. These TMSCs also had higher gene and protein expression of angiogenic molecules, as shown by microarray, qRT-PCR, and Western blotting analyses. Although the TMSCs showed the capacity to establish a capillary-like network, the structural integrity of the network was much weaker than those from HUVECs, dissipating after a few hours.

During the process of optimizing EC differentiation, our study revealed that the time required to establish the capillary-like network depended largely on the time of preconditioning. Apparently, 4 days of preconditioning provides the optimal priming step for TMSCs to effectively enter into the EC lineage; 7 days of preconditioning formed cotton thread-like structure, whereas 4 days of preconditioning resulted in the capillary-like network closely resembled to those observed with HUVECs. Passaging the cells initiates a primed signaling state in human pluripotent stem cells [36], and a similar event might have occurred with the TMSCs after 4 days of preconditioning. In contrast, a longer duration of preconditioning, more than 7 days, could negate that ideal priming process. We have no clear explanation for this phenomenon, but the increased in vitro aging of cells from a long period of cell expansion could cause them to lose their differentiation potential [37, 38]. It has been reported that long-term serial passage and/or maintenance of cells eventually causes them to lose their potential to differentiate into other cell types [36, 39, 40]. In this regard, our previous study demonstrated a reduction in differentiation markers with increasing passages and a passage-dependent decrease in the adipogenic and chondrogenic differentiation potential of TMSCs, even though the osteogenic differentiation potential increased to passage 10 and decreased thereafter [16]. Therefore, the dramatic shortening of the time required for EC differentiation could partially be explained by cellular senescence to differentiation signals in TMSCs during 7 days of expansion.

We also observed that, compared with DMEM preconditioning, EGM preconditioning produced capillary-like networks with higher stability, which resembled more closely with those observed in HUVECs. Because EGM is specifically designed for the growth and maintenance of ECs [41] and has been used extensively to differentiate MSCs into EC-like cells in previous studies [11, 29], it is sensible that TMSCs in EGM constructed much better capillary-like networks than TMSCs in DMEM.

In our genome analyses to investigate how EGM could have affected the EC differentiation of TMSCs in more detail, the most distinctive changes occurred in genes that are functionally related to vasculature development and angiogenesis: HIF1α, IL8, ANGPT1, ENPP2, and IGFBP5. It is unclear exactly how EGM preconditioning induces these gene expression changes, but they seem to be associated with the increased phosphorylation of the transcription molecule Akt-Ser⁴⁷³, an activation marker of the Phosphatidylinositol-3-kinase (PI3K)/Akt pathway, as shown in the Western blotting analyses. In this regard, activated PI3K/Akt signaling was reported to induce the expression of HIF1α, eNOS and vWF to support the angiogenesis and vascular development of EC-like cells [42–44], as evident from the genome analysis, Western blotting analysis, and IFA in this study. We also observed lower gene expression of IGFBP5, an anti-angiogenic protein, in TMSCs in EGM compared with those in DMEM [45], suggesting further enhanced angiogenesis and likely formation of a capillary-like network by TMSCs. In fact, EGM stimulated the expression of eNOS and vWF protein as well as capillary-like network formation by TMSCs, which were not observed in the TMSCs in DMEM (Figs. 6, 7). Babaei and Stewart (2002) previously reported that co-culturing with smooth muscle cells with eNOS overexpression induced the angiogenesis of ECs, thereby promoting capillary-like network formation [46]. Considering that EGM preconditioning resulted in weak but clear eNOS and higher vWF expression as well as a capillary-like network that resembled those observed in HUVECs, eNOS induced by the activated Akt signaling pathway is probably at least partially responsible for the ability of the preconditioned TMSCs in EGM to establish a capillary-like network. However, further investigation is required to find the detailed mechanisms.

In this study, p-Akt-Ser⁴⁷³ expression was not visually observed for the basal HUVECs. Akt is known to be phosphorylated (in a form of p-Akt-Ser⁴⁷³) and activated by several stimulators, such as CCN1 [28] and VEGF [47], and therefore, a low amount (5 μg) of the protein lysate from the HUVECs without a simulation was probably not sufficient enough for us to visibly observe the p-Akt-Ser⁴⁷³ using Western blotting analyses in this study.

Having found that supplementation with angiogenic cytokines/factors had minimal effect on capillary-like network formation by TMSCs, we tried to mimic the culture conditions of HUVECs to see whether it could improve network formation. The incorporation of a gelatin-coated surface during preconditioning improved the integrity of the capillary-like network formed by TMSCs preconditioned in EGM, and the network structure was maintained for a longer period (up to 10 h). Furthermore, the capillary-like network was much more morphologically similar to those observed in HUVECs than with the other conditions we tested. Recent evidence suggests that stem cell differentiation can be significantly influenced by the physical properties of gelatin, such as matrix stiffness [20] and porosity [48]. The different stiffnesses of matrices using gelatin methacrylate hydrogel influenced the EC differentiation of BM-MSCs and the subsequent capillary-like structure formation. We did not test the effect of matrix stiffness and pore size on cell differentiation in this study, but adjusting those physical properties of the gelatin would probably have further promoted network formation. The mechanism by which gelatin induces differentiation potential is not clearly understood, but it might alter the 3D-conformation of adhesion or binding motifs/receptors, thereby increasing cells’ sensitivity to the microenvironment created by gelatin [49].

In this study, we sought to determine optimal conditions for TMSCs to differentiate into EC-like cells, thereby generating a novel source for treating endothelial dysfunction. It is difficult to conclude that TMSCs have the potential to differentiate into fully functional EC-like cells at this point because the expression of eNOS was much lower than that found in HUVECs; eNOS expression was visible only with the Femto ECL solution, not with the standard ECL. Additionally, the capillary-like network structure formed by the differentiated TMSCs was lasted around 10 h, whereas HUVECs’ network maintained for 20 h. For these reasons, TMSCs might not exhibit the full EC-functional potential of HUVECs at these differentiation conditions. However, the results from our study indicate that TMSCs could differentiate into EC-like cells: they established a capillary-like network with increased angiogenesis, eNOS and vWF production possibly via activation of the PI3K/Akt signaling pathways under optimal microenvironments, i.e., EGM preconditioning with gelatin coating followed by Matrigel culturing, were provided.

Taking all our results together, our study suggests that TMSCs have the potential to differentiate into EC-like cells when they are exposed to an EC-like physicochemical microenvironment, such as preconditioning with EGM and a gelatin coating. Therefore, TMSCs could be a potential source for EC regeneration to treat endothelial dysfunction.

Compliance with ethical standards

Conflicts of interest

The authors declare that they have no conflict of interest.

Ethical statement

The study protocol was approved by the Institutional Review Board of Ewha Womans University Medical Center (IRB No. ECT-11-53-02). Informed consent was confirmed by the IRB.